Most readers agree that after many years of the paper industry enjoying increased demand for paper due to blossoming computer use, the tide turned around 2000, and the demand of communications grades of paper, particularly newsprint, is in a steady decline.
Simultaneously, giant datacenters with thousands of hard discs are proliferating. There seems to be a consensus that data center energy consumption is in the order of 30 MW, and that this will rise over the next several years. Environmental advocacy organizations are increasingly criticising the data centers, mostly because of their energy consumption, and resultant increase in greenhouse gas emissions.
This situation presents an opportunity to some of our dying pulp and/or paper mills, particularly the more remote mills situated in Canada, where rivers and air are cool and clean all year round.
Data centers vary in size of course, just as paper mills do. Typically the data centers consume a few tens of MegaWatts, so are similar in power demand to many paper mills.
Data centers generally require no heat, but have significant cooling loads. The literature suggests that 24% to 50% of total power demand is for cooling. This normally requires chilling water by mechanical refrigeration equipment and cooling towers for data centers located in the US and other temperate-to-warm climates.
A 500 tpd newsprint mill using a mixture of TMP and recycled paper would typically consume about 25 MW of electric power, and have a supply of at least 10 million gallons/day of process and cooling water. In most newsprint mills this water is available from a clean source, and if not, is already treated to a fairly good level. In most parts of Canada the water is available at below 45 degrees F. This is cool enough to cool air to appropriate temperatures for data centers, even where the owners wish to minimise the entry of external air to exclude air contaminants without expensive mechanical refrigeration.
If a data center were installed to use the available power supply, benefiting from the lack of need for chillers, the energy to be dumped would raise this typical mill’s water supply by only about 35 degrees F.
Thus the data center would save electrical energy costs, as well as the capital and operating costs for a large part of the normally required water chillers and cooling towers.
My own rough energy balance above suggests that Northern mills could consider conversion to data centers as an alternative to shutting down. This is not a new idea. Google Inc. purchased the former Stora Enso mill at Hamila in Finland in 2009, and installed a data center in 2011. Presumably it is successful, since Google Inc announced in August that they are investing nearly $200 million to expand it. Employment will reportedly be increased from 90 to 115. Click on this link for some details, including how the old machine room is being used.
While some of the jobs in a data center require degrees in computer science and related fields, many are within reach of skilled maintenance and operations personnel accustomed to 24 hour shift work in an industry where reliability is important.
The more public spirited / highly visible data center owners (Such as Facebook, Google and Microsoft) are currently trying to green their image. Many mills have small hydro electric and/or biomass power generating facilities that would be compatible with the data center’s needs.
While most data centers are located in traditional industrial and semi-industrialised regions, the industry has shown interest in remote locations, particularly when there are in politically secure regions with low earthquake risk, such as Eastern Canada and Scandinavia.
Facebook is building a large data center in Lulea, on the Arctic Circle, in Sweden. This will reportedly be cooled entirely by outdoor air, without even needing the cold water mentioned above. An engineer I met this summer who had worked on the project related that political and earthquake stability were significant factors in site selection.
When a paper mill in a small, remote, town is facing shutdown, governments are often quite helpful to new industrial developments, which should make replacing a dying mill with a data center all the more attractive to a developer.
I find it hard to imagine any paper company branching out into data center development, but there seems to be a case for actively seeking out cooperation with data center builders to reduce the financial and social costs of shutting down remote mills.